基于灰色GM-BP神经网络组合模型的中国镍原矿多情景需求预测

doi:10.13776/j.cnki.resourcesindustries.20231123.001

摘要/Abstract

摘要：

2016年我国颁布《全国矿产资源规划(2016—2020年)》，首次将镍列为战略性矿产资源。我国是全球最大的镍消费国，但镍资源储量少，对外依存度高，科学预测镍原矿需求量对保障镍矿产业链与供应链安全具有重要的现实意义。从需求侧出发，利用灰色关联度法选取中国不锈钢产量、人均GDP、电镀行业市场规模、城镇化率、产业结构、新能源汽车产量作为镍原矿需求情景预测的驱动变量，再在灰色GM(1,1)模型预测基础上，与BP神经网络算法相结合，构建基于残差优化的GM-BP组合模型，对2025—2035年中国镍原矿需求展开多情景预测。研究结果表明：组合模型实现了对小样本非线性时间序列数据的有效预测，且比GM(1,1)模型拟合误差更小，预测精度更高；根据组合模型，2025年、2030年、2035年我国镍原矿多情景需求均值分别为182.22万t、272.08万t、395.17万t，“十四五”“十五五”“十六五”期间需求年均增长4.26%、10.54%、9.78%。镍原矿需求呈稳定上升态势，镍矿供需矛盾将进一步加剧，我国必须提高镍供应能力，降低对进口镍的依赖程度。对此，提出如下政策建议：1)推进国内不锈钢产业的转型升级，优化生产工艺和产品结构，推广新型合金材料的应用；2)加大对镍矿勘探和开发的支持力度，如鼓励矿业企业技术创新，提高勘探效率和精度，同时积极推动国际合作，吸引国外先进技术、设备进入国内市场；3)促进进口多元化，与多个供应国建立合作关系，鼓励国内企业参与海外镍矿项目。

关键词: GM-BP模型, BP神经网络, 镍原矿需求, 情景预测

Abstract: National Mineral Resources Planning (2016-2020), issued in 2016, lists nickel as strategic mineral. China is the largest nickel consumer, but insufficient in nickel resources with high dependence on external supply. Scientific forecast of raw nickel demand is of significance to secure nickel production and supply chains. This paper uses gray correlation to select China’s stainless steel production, GDP per capita, electroplating market scale, urbanization rate, industrial structure, and new energy vehicle production as driving variables to forecast nickel demand under different scenarios from demand side, and combines gray GM(1, 1) model with BP neural network to construct a GM-BP model based on residual minimization, which is used to forecast China’s nickel demand from 2015 to 2035. This compound model is effective in forecasting non-linear sequence data, with the fitting error smaller than GM(1, 1) model. China’s demand for raw nickel is forecasted at 1 822.2 kt in 2025, 2 720.8 kt in 2030 and 3 951.7 kt in 2035 at an annual rising rate of 4.26% in the 14^th Five-Year Plan, 10.54% in the 15^th Five-Year Plan and 9.78% in the 16^th Five-Year Plan. The rising demand trend for raw nickel will lead to a conflict between its supply and demand. China has to raise its supply capacity and decrease dependence on imports. This paper presents suggestions on upgrading stainless steel industry, optimizing processing and producing new alloy materials, and on boosting nickel exploration and development, and on diversifying import sources, and on supporting Chinese investors’ overseas nickel projects.

Key words: GM-BP model, BM neural network, raw nickel demand, scenario forecast

中图分类号:

F205

周文潇, 詹成, 张周益, 等.

基于灰色GM-BP神经网络组合模型的中国镍原矿多情景需求预测 [J]. 资源与产业, 2024, 26(2): 53-66.

ZHOU Wenxiao, ZHAN Cheng, ZHANG Zhouyi, et al.

MULTI-SCENARIO DEMAND FORECAST OF RAW NICKEL IN CHINA BASED ON GRAY GM-BP NEURAL NETWORK COMPOUND MODEL [J]. Resources & Industries, 2024, 26(2): 53-66.

参考文献

安海忠, 李华姣, 2022. 战略性矿产资源全产业链理论和研究前沿[J]. 资源与产业, 24(1): 8-14.〔AN H Z, LI H J, 2022. Theory and research advances in whole industrial chain of strategic mineral resources[J]. Resources & Industries, 24(1): 8-14.〕

陈海涛, 2012. 基于灰色系统理论的我国石油消费对外依存度预测[J]. 生态经济(5): 103-106.〔CHEN H T, 2012. The forecast of degree of external dependence on oil consumption in China based on grey system theory[J]. Ecological Economy (5):103-106.〕

陈锡康, 杨翠红, 祝坤福, 等, 2023. 2023年中国经济增长速度的预测分析与政策建议[J]. 中国科学院院刊, 38(1): 81-90.〔CHEN X K, YANG C H, ZHU K F, et al., 2023. Forecast of China’s economic growth rate in 2023 and policy suggestions[J]. Bulletin of Chinese Academy of Sciences, 38(1): 81-90.〕

崔伟杰, 2020. 全球镍资源产业供需格局分析及预测[D]. 北京: 中国地质大学(北京).〔CUI W J, 2020. Analysis and forecast of supply and demand pattern of global nickel resource industry[D]. Beijing: China University of Geosciences (Beijing).〕

董晓红, 冯芷蔚, 张家安, 等, 2022. 含非线性残差的新能源汽车规模预测方法[J]. 电力工程技术, 41(5): 76-84.〔DONG X H, FENG Z Y, ZHANG J A, et al., 2022. New energy vehicle scale prediction method with nonlinear residuals[J]. Electric Power Engineering Technology, 41(5): 76-84.〕

杜修立, 张昱昭, 2022. 中国城镇化率提升的动力分解与新发展阶段趋势预测: 基于国际比较的一种新方法[J]. 统计研究, 39(2): 33-47.〔DU X L, ZHANG Y Z, 2022. Driving-force decomposition increase of China’s urbanization rate increase and trend forecast in the new development stage: a new method based on international comparison[J]. Statistical Research, 39(2): 33-47.〕

高春亮, 魏后凯, 2013. 中国城镇化趋势预测研究[J]. 当代经济科学, 35(4): 85-90.〔GAO C L, WEI H K, 2013. Prediction study on the urbanization trends of China[J]. Modern Economic Science, 35(4): 85-90.〕

高芯蕊, 王安建, 2010. 基于“S”规律的中国钢需求预测[J]. 地球学报, 31(5): 645-652.〔GAO X R, WANG A J, 2010. The prediction of China’s steel demand based on S-shaped regularity[J]. Acta Geoscientica Sinica, 31(5): 645-652.〕

郭克莎, 彭继宗, 2022. 二三产业结构变动与经济发展质量: 上中等收入阶段向高收入阶段演进的国际经验[J]. 财贸经济, 43(8): 5-26.〔GUO K S, PENG J Z, 2022. The relationship between structural changes in secondary and tertiary industries and the quality of economic development: an international empirical study from the perspective of the evolution from the upper-middle income stage to the high-income stage[J]. Finance and Trade Economics, 43(8): 5-26.〕

黄建华, 张迪, 2022. 面向不确定物流需求的改进GM-BPNN组合预测方法[J]. 统计与决策, 38(16): 26-29.〔HUANG J H, ZHANG D, 2022. An improved GM-BPNN combined forecasting method for uncertain logistics demand[J]. Statistics and Decision, 38(16): 26-29.〕

鞠建华, 张照志, 潘昭帅, 等, 2022. 我国战略性新兴产业矿产厘定与“十四五”需求分析[J]. 中国矿业, 31(9): 1-11.〔JU J H, ZHANG Z Z, PAN Z S, et al., 2022. Determination of mineral resources in China’ s strategic emerging industries and analysis of the demand of the“14th five year plan”[J]. China Mining Magazine, 31(9): 1-11.〕

李代红, 2014. 基于BP神经网络和GM(1,1)的我国三次产业预测[J]. 统计与决策(7): 81-83.〔LI D H, 2014. China’s three industry forecasts based on BP neural network and GM(1,1)[J]. Statistics and Decision (7): 81-83.〕

李剑波, 鲜学福, 2016. 基于灰色神经网络模型的重庆能源需求预测[J]. 西南大学学报(自然科学版), 38(6): 136-141.〔LI J B, XIAN X F, 2016. Energy demand forecasting of Chongqing based on grey neural network model[J]. Journal of Southwest University(Natural Science), 38(6): 136-141.〕

刘连义, 刘思峰, 吴利丰, 2024. 基于离散时间灰色幂模型的新能源汽车销售量预测[J]. 中国管理科学, 32(1): 106-114.〔LIU L Y, LIU S F, WU L F, 2024. New energy vehicle sales forecast based on Siscrete grey power model[J]. Chinese Journal of Management Science, 32(1): 106-114.〕

那丹妮, 王高尚, 2010. 全球镍需求趋势预测[J]. 资源与产业, 12(6): 53-57.〔NA D N, WANG G S, 2010. Forecast in global nickel demand trend[J]. Resources & Industries,12(6): 53-57.〕

彭乃驰, 党婷, 2016. 基于ARMA-GM-BP组合预测模型及应用[J]. 统计与决策(2): 80-82.〔PENG N C, DANG T, 2016. Based on ARMA-GM-BP combined prediction model and application[J]. Statistics and Decision (2): 80-82.〕

清华大学中国经济思想与实践研究院ACCEPT宏观预测课题组, 2023. 重振增长〖KG*2〗释放活力: 2023—2027年中国经济发展展望[J]. 改革(1): 31-50.〔ACCEPT, 2023. Reinvigorate growth and release vitality: China’s economic development outlook of 2023-2027[J]. Reform (1): 31-50.〕

任英欣, 刘金伟, 2022. 环境保护法视角下电镀行业发展策略研究[J]. 电镀与精饰, 44(11): 107-108.〔REN Y X, LIU J W, 2022. Research on the development strategy of electroplating industry from the perspective of environmental protection law[J]. Plating and Finishing, 44(11): 107-108.〕

史育龙, 郭巍, 2022. 高质量推进我国城镇化与碳达峰的国际经验镜鉴: 基于OECD数据考察[J]. 生态经济, 38(4): 29-34.〔SHI Y L, GUO W, 2022. International experience of promoting China’s urbanization and carbon peak with high quality: based on OECD data[J]. Ecological Economy, 38(4): 29-34.〕

王安建, 高芯蕊, 2020. 中国能源与重要矿产资源需求展望[J]. 中国科学院院刊, 35(3): 338-344.〔WANG A J, GAO X R, 2020. China’s energy and important mineral resources demand perspective[J]. Bulletin of the Chinese Academy of Sciences, 35(3): 338-344.〕

王保贤, 刘毅, 2018. 基于灰色BP神经网络模型的人力资源需求预测方法[J]. 统计与决策, 34(16): 181-184.〔WANG B X, LIU Y, 2018. Human resource demand forecasting method based on gray BP neural network model[J]. Statistics and Decision, 34(16): 181-184.〕

王欢, 马冰, 贾凌霄, 等, 2021. 碳中和目标下关键矿产在清洁能源转型中的作用、供需分析及其建议[J]. 中国地质, 48(6): 1720-1733.〔WANG H, MA B, JIA L X, et al., 2021. The role, supply and demand of critical minerals in the clean energy transition under carbon neutrality targets and their recommendations[J]. Geology in China, 48(6): 1720-1733.〕

王修, 李天骄, 王安建, 等, 2022. 基于主要工业产品产量的我国钽资源需求预测[J]. 矿业研究与开发, 42(6): 191-196.〔WANG X, LI T J, WANG A J, et al., 2022. The forecast of tantalum resource demand based on the output of major industrial products in China[J]. Mining Research and Development, 42(6): 191-196.〕

闻少博, 陈志华, 刘雪勇, 2022. 中美博弈视角下中国稀土资源供应风险研究[J]. 资源与产业, 24(5): 1-9.〔WEN S B, CHEN Z H, LIU X Y, 2022. China’s REE supply risks viewing from China-America’s competition[J]. Resources & Industries, 24(5): 1-9.〕吴振信, 帅加琴, 王书平, 2014. 基于改进GM(1,1)模型的中国铜消费需求预测[J]. 工业技术经济, 33(8): 9-14.〔WU Z X, SHUAI J Q, WANG S P, 2014. Forecasting of Chinese copper demand based on improved gray model[J]. Journal of Industrial Technological Economics, 33(8): 9-14.〕

邢佳韵, 张晓鹤, 陈其慎, 等, 2021. “二元消费”影响下的镍供需形势分析[J]. 地球学报, 42(2): 251-257.〔XING J Y, ZHANG X H, CHEN Q S, et al., 2021. An analysis of nickel supply and demand situation under the influence of “dual consumption”[J]. Acta Geoscientica Sinica, 42(2): 251-257.〕

徐爱东, 陈瑞瑞, 李烁, 等, 2021. 镍钴行业发展形势分析及建议[J]. 中国有色冶金, 50(6): 9-15.〔XU A D, CHEN R R, LI S, et al., 2021. Analysis and suggestions on the development of nickel-cobalt industry[J]. China Nonferrous Metallurgy, 50(6): 9-15.〕

徐奇栋, 吕启全, 廖福源, 2016. 基于灰色系统理论对矿产资源需求的预测研究: 以浙江省为例[J]. 科技通报, 32(10): 17-20.〔XU Q D, LÜ Q Q, LIAO F Y, 2016. Forecast and study on the minerals resources demand based on the grey system theory: taking Zhejiang province as an example[J]. Bulletin of Science and Technology, 32(10): 17-20.〕

杨俊峰, 潘寻, 2021. “十四五”中国锂动力电池产业关键资源供需分析[J]. 有色金属(冶炼部分)(6): 37-41.〔YANG J F, PAN X, 2021. Analysis on supply and demand of key resources of lithium power battery industry in China during the 14th Five-Year Plan period[J]. Nonferrous Metals(Extractive Metallurgy) (6): 37-41.〕

尹伟华, 2021. “十四五”时期我国产业结构变动特征及趋势展望[J]. 中国物价(9): 3-6.〔YIN W H, 2021. The characteristics and prospects industrial structure in China during the 14th Five-Year Plan[J]. China Price (9): 3-6.〕

张俊深, 袁程炜, 2016. 基于BP神经网络与修正GM(1,1)模型的能源消费组合预测[J]. 统计与决策(5): 90-93.〔ZHANG J S, YUAN C W, 2016. Energy consumption combination prediction based on BP neural network and modified GM(1,1) model[J]. Statistics and Decision (5): 90-93.〕

张明喜, 丛树海, 2009. 我国财政风险非线性预警系统: 基于BP神经网络的研究[J]. 经济管理, 31(5): 147-153.〔ZHANG M X, CONG S H, 2009. Nonlinear early warning system of financial risk in China: research based on BP neural network[J]. Economic Management Journal, 31(5): 147-153.〕

张鹏, 2020. 小样本时间序列灰色预测关键技术研究[D]. 成都: 电子科技大学.〔ZHANG P, 2020. Research on key techniques of grey prediction for small sample time series[D]. Chengdu: University of Electronic Science and Technology.〕

张晓晶, 2022. 我国中长期增长的预测、挑战与应对[J]. 国外社会科学(5): 36-42.〔ZHANG X J, 2022. China’s medium and long-term growth: forecasts, challenges and responses[J]. Social Sciences International (5): 36-42.〕

张泽南, 张照志, 吴晴, 等, 2020. 中国锂矿资源需求预测[J]. 中国矿业, 29(7): 9-15.〔ZHANG Z N, ZHANG Z Z, WU Q, et al., 2020. Chinese lithium mineral resource demand forecast[J]. China Mining Magazine, 29(7): 9-15.〕

郑明贵, 于明, 范秋蓉, 等, 2023. 中国2025—2035年碳酸锂需求预测: 基于灰色关联分析和ARIMA-GM-BP神经网络的组合模型[J]. 地球科学进展, 38(4): 377-387.〔ZHENG M G, YU M, FAN Q R, et al., 2023. China’s lithium carbonate demand forecast 2025—2035: a combined model based on grey correlation analysis and the ARIMA-GM-BP neural network[J]. Advances in Earth Science, 38(4): 377-387.〕

中国社会科学院宏观经济研究中心课题组, 李雪松, 陆旸, 等, 2020. 未来15年中国经济增长潜力与“十四五”时期经济社会发展主要目标及指标研究[J]. 中国工业经济(4): 5-22.〔RESEARCH GROUP OF THE MACROECONOMIC RESEARCH CENTER OF THE CASS, LI X S, LU Y, et al., 2020. Research on the potential growth of the chinese economy in the next 15 years and the main goals and indicators of economic and social development during the 14th Five-Year Plan period[J]. China Industrial Economics (4): 5-22.〕

中国有色金属报, 2021. 电动车用镍需求旺盛[J]. 中国有色冶金(4): 78.〔CHINA NONFERROUS METALS NEWS, 2021. Nickel for electric vehicles is in high demand[J]. China Nonferrous Metallurgy (4): 78.〕

中华人民共和国自然资源部, 2016. 全国矿产资源规划(2016—2020年)[EB/OL]. (2016-11-15)[2023-05-06]. http://g. mnr. gov. cn/201701/t20170123_1430456. html.〔Ministry of Natural Resources of the People’s Republic of China, 2016. National mineral resources planning (2016—2020)[EB/OL]. (2016-11-15)[2023-05-06]. http://g. mnr. gov. cn/201701/t20170123_1430456. html.〕

中华人民共和国自然资源部, 2022. 中国矿产资源报告2022[M]. 北京: 地质出版社.〔Ministry of Natural Resources of the People’s Republic of China, 2022. China mineral resources report 2022[M]. Beijing: Geology Press.〕

周文浩, 曾波, 2020. 灰色关联度模型研究综述[J]. 统计与决策, 36(15): 29-34.〔ZHOU W H, ZENG B, 2020. A research review of grey relational degree model[J]. Statistics and Decision, 36(15): 29-34.〕

周扬, 吴文祥, 胡莹, 等, 2010. 基于组合模型的能源需求预测[J]. 中国人口·资源与环境, 20(4): 63-68.〔ZHOU Y, WU W X, HU Y, et al., 2010. Energy demand forecasting based on combined model[J]. China Population, Resources and Environment,20(4):63-68.〕

朱强, 2010. 基于灰色系统理论的经济建模方法[J]. 统计与决策(11): 27-28.〔ZHU Q, 2010. Economic modeling methods based on grey systems theory[J]. Statistics and Decision (11): 27-28.〕

朱晓宵, 刘明, 曹杰, 2023. 基于滚动灰色GM(1,1)修正的新冠肺炎疫情演化情景重建[J]. 运筹与管理, 32(10): 95-101.〔ZHU X X, LIU M, CAO J, 2023. Reconstruction of COVID-19 epidemic scenario: a modified model based on rolling grey GM(1,1)[J]. Operations Research and Management Science, 32(10): 95-101.〕

BAHMANI M, NEJATI M, GHASEMINEJAD A, et al., 2021. A novel hybrid approach based on BAT algorithm with artificial neural network to forecast Iran’s oil consumption[J]. Mathematical Problems in Engineering (8): 1-9.

ESCAVY J I, HERRERO M J, TRIGOS L, et al., 2020. Demographic vs economic variables in the modelling and forecasting of the demand of aggregates: the case of the Spanish market (1995-2016)[J]. Resources Policy, 65(9):101537. DOI:10. 1016/j. resourpol. 2019. 101537.

GALOS K, LEWICKA E D, KAMYK J, et al., 2021. Forecast trends in demand for deficit key minerals for the Polish economy[J]. Gospodarka Surowcami Mineralnymi-Mineral Resources Management, 37(3): 5-29.

LI B L, BICKNELL K B, RENWICK A, 2019. Peak phosphorus, demand trends and implications for the sustainable management of phosphorus in China[J]. Resources, Conservation and Recycling, 146: 316-328.

MA M L, WANG Z Z, 2019. Prediction of the energy consumption variation trend in South Africa based on ARIMA, NGM and NGM-ARIMA models[J]. Energies, 13(1): 10.

PUENTES J A, RIBEIRO C O, RUELAS E A, et al., 2021. Ethanol fuel demand forecasting in Brazil using a LSTM recurrent neural network approach[J]. IEEE Latin America Transactions, 19(4): 551-558.

REN M, DAI J Y, ZHU W C, et al., 2021. Combined modelling for iron ore demand forecasting with intelligent optimization algorithms[J]. Gospodarka Surowcami Mineralnymi-Mineral Resources Management, 37(1): 21-38.

SVERDRUP H U, OLAFSDOTTIR A H, 2019. Assessing the long-term global sustainability of the production and supply for stainless steel[J]. Biophysical Economics and Resource Quality, 4(2): 8.

[1]	高珊, 曹明霞, 李丹. 江苏省碳排放特征分析及达峰预测[J]. 资源与产业, 2023, 25(3): 1-9.
[2]	杜焱, 胡鑫杨. 我国2030年实现碳达峰路径研究——基于经济、能源、碳排放系统的SD模型[J]. 资源与产业, 2022, 24(5): 19-28.
[3]	王利军, 庞雅倩, 陈梦冬. 湖北省交通业碳排放影响因素及情景预测[J]. 资源与产业, 2022, 24(3): 106-113.
[4]	康凯,宋文博，郭子龙，孟庆香. 河南省环境污染和经济增长关系实证研究:基于遗传—BP神经网络、协整分析和VAR模型[J]. 资源与产业, 2014, 16(5): 100-106.
[5]	陈家愿，郑明贵. 基于BP神经网络的我国海外矿业投资金融风险预警分析[J]. 资源与产业, 2014, 16(4): 106-110.
[6]	袁纬芳，郑明贵. 基于BP神经网络的我国铝矿资源需求情景分析[J]. 资源与产业, 2014, 16(1): 132-137.
[7]	石艳丽, 安海忠, 高湘昀. 基于时序—神经网络模型的我国石油消费预测[J]. 资源与产业, 2011, 13(2): 37-42.
[8]	常胜, 李江风. 基于BP神经网络的城市土地集约利用评价——以湖北省鄂州为例[J]. 资源与产业, 2010, 12(2): 31-34.